Israeli researchers have announced that they have invented a computer algorithm that can decipher ancient texts. This may probably be the basis of a search engine dedicated exclusively to ancient texts.
The application makes use of pattern recognition similar to the fingerprint identification. However, in this case the program recognizes letters, words and writing styles. This software will save historians a lot of time by making analysis of manuscripts much faster. Analysis of ancient texts usually takes many years. However, using the software, it is now possible to analyze them in a matter of minutes.
This software will be especially useful in analyzing historical documents that have faded over time. The software learns from experience, so it will become smarter and more accurate as it analyzes more texts.
The software application works with digital copies of the texts. Each pixel of writing is assigned a number value depending on how dark it is. The software separates the text from the background and then identifies individual characters of the text. It is capable of analyzing handwriting and writing styles and can fill in smeared or faded characters. If faded texts were to be analyzed by hand, it would be very difficult and time consuming.
The research team behind this application is currently focusing on ancient Hebrew texts. However the software can also be used with other languages such as Sanskrit.
The details of the algorithm and software were published in the online version of “Pattern Recognition”. They will be published on paper in December in the same journal.
The researchers expressed hope that a general purpose version of their software will become available within the next two years. Bar-Yosef, the leader of the research team said, “A program for all academics could be ready in two years.”
As libraries in many countries try to digitize their collections, software such as this can be very beneficial in creating digital databases of handwritten texts.